Electroencephalography (EEG) is a crucial tool for recording the brain’s electrical activity, providing insights into neural processes. However, extracting meaningful information of a complex visual stimulus (e.g. a scene image) from EEG data is challenging due to a large domain shift and the signal’s complexity. Recent research efforts focus on decoding visual information from EEG using advanced models and techniques. In our work, we address structure and locality estimation from EEG signals with a particular focus on depth perception, which is a fundamental aspect of the visual processing pathway and could serve as a key intermediate ground for visual decoding models. Thus, we focus on the task of reconstruction of depth maps from EEG data, corresponding to the images shown to subjects. Our work involves a contrastive learning framework in an attempt to align GNN based EEG embeddings to that of and depth map embeddings. We also perform experiments to draw some insights about the importance of EEG channels in such a EEG to Depth map reconstruction.